Publicação

Leveraging Computer Vision to Enhance Waste Management Education in Portugal: Practical Considerations in Developing a Comprehensive Method for Waste Recognition

Ver documento

Detalhes bibliográficos
Resumo:This thesis presents the development of a computer vision model aimed at the future implementation of ReciclAR, a mobile application designed to facilitate recycling through advanced object detection technology. The research emphasizes creating a comprehensive dataset tailored to training computer vision models to recognize diverse waste materials in Portuguese household contexts. By using the YOLO framework, the model is trained to identify recyclable items, aiming to enhance user engagement and promote sustainable practices. The study involves a detailed performance comparison of YOLOv8 and YOLOv10 models and sizes, on our custom datasets. Additionally, this thesis discusses the design of a mockup mobile application incorporating Human-Computer Interaction principles to ensure an intuitive user experience. This research aims to reduce wishcycling and bin waste contamination by accurately identifying recyclable materials, bridging the gap between policy targets and household recycling practices, and contributing to a sustainable future.
Autores principais:Rocha, Inês Leónidas Xavier Esteves Ferreira da
Assunto:Computer Vision Deep Learning Object Detection Sustainability Recycling YOLO SDG 11 - Sustainable cities and communities SDG 12 - Responsible production and consumption
Ano:2024
País:Portugal
Tipo de documento:dissertação de mestrado
Tipo de acesso:acesso aberto
Instituição associada:Universidade Nova de Lisboa
Idioma:inglês
Origem:Repositório Institucional da UNL
Descrição
Resumo:This thesis presents the development of a computer vision model aimed at the future implementation of ReciclAR, a mobile application designed to facilitate recycling through advanced object detection technology. The research emphasizes creating a comprehensive dataset tailored to training computer vision models to recognize diverse waste materials in Portuguese household contexts. By using the YOLO framework, the model is trained to identify recyclable items, aiming to enhance user engagement and promote sustainable practices. The study involves a detailed performance comparison of YOLOv8 and YOLOv10 models and sizes, on our custom datasets. Additionally, this thesis discusses the design of a mockup mobile application incorporating Human-Computer Interaction principles to ensure an intuitive user experience. This research aims to reduce wishcycling and bin waste contamination by accurately identifying recyclable materials, bridging the gap between policy targets and household recycling practices, and contributing to a sustainable future.